Toward Atomistic Resolution Structure of ... - ACS Publications

Oct 28, 2015 - open collaboration, using the nmrlipids.blogspot.fi blog61 as a communication platform. Our approach is inspired by the. Polymath proje...
1 downloads 7 Views 1MB Size
Subscriber access provided by UNIV OF CAMBRIDGE

Article

Towards Atomistic Resolution Structure of Phosphatidylcholine Headgroup and Glycerol Backbone at Different Ambient Conditions Alexandru Botan, Favela Fernando, Patrick François,Jimi Fuchs, Matti Javanainen, Matej Kanduc, Waldemar Kulig, Antti Lamberg, Claire Loison, Alexander P. Lyubartsev, Markus Sakari Miettinen, Luca Monticelli, Jukka Määttä, Samuli OH Ollila, Marius Retegan, Tomasz Róg, Hubert Santuz, and Joona Petteri Tynkkynen J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.5b04878 • Publication Date (Web): 28 Oct 2015 Downloaded from http://pubs.acs.org on November 2, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry B is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

1

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Towards Atomistic Resolution Structure of Phosphatidylcholine Headgroup and Glycerol Backbone at Different Ambient Conditions† Alexandru Botan,‡ Fernando Favela-Rosales,¶ Patrick F. J. Fuchs,§ Matti Javanainen,k Matej Kanduˇc,⊥ Waldemar Kulig,k Antti Lamberg,# Claire Loison,‡ Alexander Lyubartsev,@ Markus S. Miettinen,△ Luca Monticelli,∇ Jukka M¨a¨att¨a,†† O. H. Samuli Ollila,∗,†† Marius Retegan,‡‡ Tomasz Rog,k Hubert Santuz,¶¶,§§,kk,⊥⊥ and Joona Tynkkynenk Institut Lumi`ere Mati`ere, UMR5306 Universit´e Lyon 1-CNRS, Universit´e de Lyon 69622 Villeurbanne, France, Departamento de F´ısica, Centro de Investigaci´ on y de Estudios Avanzados del IPN, Apartado Postal 14-740, 07000 M´exico D.F., M´exico, Institut Jacques Monod, CNRS, Universit´e Paris Diderot, Sorbonne Paris Cit´e, Paris, France, Department of Physics, Tampere University of Technology, Tampere, Finland, Fachbereich Physik, Freie Universitat Berlin, Berlin, Germany, Department of Chemical Engineering, Kyoto University, Kyoto, Japan, Division of Physical Chemistry, Department of Materials and Environmental Chemistry, Stockholm University, S-106 91 Stockholm, SWEDEN, Fachbereich Physik, Freie Universit¨ at Berlin, Berlin, Germany, Institut de Biologie et Chimie des Prot´eines (IBCP), CNRS UMR 5086, Lyon, France, Aalto University, Espoo, Finland, Max Planck Institute for Chemical Energy Conversion, Mulheim an der Ruhr, Germany, INSERM, UMR S 1134, DSIMB, Paris, France, Universit´e Paris Diderot, Sorbonne Paris Cit´e, UMR S 1134, Paris, France, Institut National de la Transfusion Sanguine (INTS), Paris, France, and Laboratoire d’Excellence GR-Ex, Paris, France E-mail: [email protected]. 2

ACS Paragon Plus Environment

Page 2 of 48

Page 3 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Abstract Phospholipids are essential building blocks of biological membranes. Despite of vast amount of very accurate experimental data, the atomistic resolution structures sampled by the glycerol backbone and choline headgroup in phoshatidylcholine bilayers are not known. Atomistic resolution molecular dynamics simulations have the potential to resolve the structures, and to give an arrestingly intuitive interpretation of the experimental data—but only if the simulations reproduce the data within experimental accuracy. In the present work, we simulated phosphatidylcholine (PC) lipid bilayers with 13 different atomistic models, and compared simulations with NMR experiments in terms of the highly structurally sensitive C–H bond vector order parameters. Focusing on the glycerol backbone and choline headgroups, we showed that the order parameter comparison can be used to judge the atomistic resolution structural accuracy of the models. Accurate models, in turn, allow molecular dynamics simulations to be used as an interpretation tool that translates these NMR data into a dynamic three dimensional representation of biomolecules in biologically relevant conditions. In addition to lipid bilayers in fully hydrated conditions, we reviewed previous experimental data for dehydrated bilayers and cholesterol-containing bilayers, and interpreted them with simulations. Although none of the existing models reached experimental accuracy, by critically comparing them we were able to distill relevant chemical information: (1) increase of choline †

Publication about results presented in the NMRlipids project To whom correspondence should be addressed ‡ Lyon CNRS ¶ Mexico § CNRS Paris k Tampere University of Technology ⊥ Freie Universit¨at Berlin # Kyoto University @ Stockholm University △ Freie Universit¨at Berlin ∇ IBCP †† Aalto University ‡‡ Max Planck ¶¶ INSERM §§ Diderot kk INTS ⊥⊥ Labex ∗

3

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

order parameters indicates the P–N vector tilting more parallel to the membrane, and (2) cholesterol induces only minor changes to the PC (glycerol backbone) structure. This work has been done as a fully open collaboration, using nmrlipids.blogspot.fi as a communication platform; all the scientific contributions were made publicly on this blog. During the open research process, the repository holding our simulation trajectories and files (https://zenodo.org/collection/user-nmrlipids) has become the most extensive publicly available collection of molecular dynamics simulation trajectories of lipid bilayers.

Keywords lipid, bilayer, molecular dynamics simulation, nuclear magnetic resonance, order parameter

Introduction Phospholipids containing various polar headgroups and acyl chains are essential building blocks of biological membranes. Lamellar phospholipid bilayer structures have been widely studied with various experimental and theoretical techniques as a simple model for cellular membranes. 1–8 Phospholipid molecules are composed of hydrophobic acyl chains connected by a glycerol backbone to a hydrophilic headgroup; see Fig. 1 for the structure of 1-palmitoyl2-oleoylphosphatidylcholine (POPC). The behaviour of the acyl chains in a lipid bilayer is relatively well understood. 1–5,8,9 The conformations sampled by the glycerol backbone and choline in a fluid bilayer are, however, not fully resolved as even the most accurate scattering and Nuclear Magnetic Resonance (NMR) techniques give only a set of values that the structure has to fulfill, but there is no unique way to derive the actual structure from them. 9–18 Some structural details have been extracted from crystal structures, 1 H NMR studies, and Raman spectroscopy, 19–25 but general consensus concerning the structures sampled in the fluid state has not been reached. 9–18,24,25 Importantly, the structural parameters

4

ACS Paragon Plus Environment

Page 4 of 48

Page 5 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

for the glycerol backbone are similar for various biologically relevant lipid species (phosphatidylcholine (PC), phosphatidylethanolamine (PE) and phosphatidylglycerol (PG)) in various environments, 26 and the structural parameters for the choline headgroup are similar in model membranes and real cells (mouse fibroblast L-M cell). 27 Thus, resolving the PC-lipid glycerol and choline structures would be useful for understanding a wide range of different biological membranes. Classical atomistic molecular dynamics simulations have been widely used to study lipid bilayers. 2–7 As these models provide an atomistic resolution description of the whole lipid molecule, they have the potential to solve the glycerol backbone and headgroup structures. The experimental C–H bond order parameters (routinely compared between experiments and simulations for the acyl chains 2–6 ) are also known for the glycerol backbone (g1 , g2 , and g3 ) and choline (α and β) segments (see Fig. 1 for definitions) and are among the main parameters used in attempts to derive lipid structures from experimental data. 10–13,15,16,18 Notably, the structures sampled in a simulation that reproduces these parameters will automatically comprise an interpretation of the experiments. In other words, such simulations can be considered as an accurate atomistic resolution description of the behavior of lipid molecules in a bilayer. Only a few studies 28–37 have compared the glycerol backbone and choline headgroup order parameters between simulations and experiments. The main reason probably is that the existing experimental data for the glycerol backbone and choline headgroups are scattered over many publications and published in a format that is difficult to understand without some NMR expertise. In addition to the order parameters, dihedral angles for the glycerol backbone and headgroup estimated from experiments, 28,38–42 and

31

31

P chemical shift anisotropy 36

P-13 C dipolar couplings 43 have been used to assess the quality of a simulation model.

In this work, we first review the most relevant experimental data for the glycerol backbone and choline headgroup order parameters in a phosphatidylcholine lipid bilayer. Then the available atomistic resolution lipid models are carefully compared to the experimental data.

5

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The comparison reveals that the CHARMM36, 31 GAFFlipid, 33 and MacRog 37 models have the most realistic glycerol backbone and choline structures. We also compare the glycerol backbone and choline structures between the most often used (Berger-based) lipid model 44 and the best performing models, to demonstrate that by using the order parameters we can distinguish the more reasonable structures from the less reasonable ones. However, none of the current models is accurate enough to properly resolve the atomistic resolution structures. In addition to fully hydrated single component lipid bilayers, the glycerol backbone and choline order parameters have been measured under a large number of changing conditions: hydration level, 45–47 cholesterol content, 35,48 ion concentration, 49–53 temperature, 54 charged lipid content, 52,53 charged surfactant content, 55 drug molecule concentration, 30,56,57 and protein content 58,59 (listing only the publications most relevant for this work and the pioneering studies). Existence of these data allows the comparison of structural responses to varying conditions between simulations and experiments, in other words, validation of the simulation models and interpretation of the original experiments. Here we demonstrate the power of this approach in understanding the behaviour of a bilayer as a function of hydration level and cholesterol content. Choline headgroup order parameters as function of ion concentration, and their relation to the ion binding affinity, are discussed elsewhere. 60

Figure 1: Chemical structure of 1-palmitoyl-2-oleoylphosphatidylcholine (POPC).

6

ACS Paragon Plus Environment

Page 6 of 48

Page 7 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Methods Open collaboration This work has been done as a fully open collaboration, using the nmrlipids.blogspot.fi blog 61 as a communication platform. Our approach is inspired by the Polymath project, 62 however there are some essential differences. We started by publishing a manuscript 63 discussing the glycerol backbone and choline structures in a Berger-based model (the most used molecular dynamics simulation model for lipid bilayers). Simultaneously, we presented an open invitation for further contributions and discussion on the blog. All the scientific contributions were made publicly through the blog. Every contributor was offered coauthorship according to the guidelines defined in the beginning of the project; 64 the acceptance of the offer was based on authors’ self-assesment of their scientific contribution. These contributions are summarized in the Supporting Information. Almost all simulation data, including input files for reproduction and trajectories for further analysis, are collected on our CERN-hosted Zenodo file repository (https://zenodo. org/collection/user-nmrlipids). Thus, in addition to the main topic of this manuscript, we present the most extensive publicly available collection of simulation trajectories for lipid bilayers, opening up numerous possiblities for different analyses with much less effort than previously required. Further information, such as scripts, figures, and manuscript text files, are available through our GitHub repository. 65

Order parameters from experiments The order parameter of a hydrocarbon C–H vector is defined as 1 SCH = h3 cos2 θ − 1i, 2

7

ACS Paragon Plus Environment

(1)

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

where the angle brackets denote an ensemble average over the sampled conformations, and θ is the angle between the C–H bond and the membrane normal. The absolute values of order parameters can be measured by detecting quadrupolar splitting with 2 H NMR 66 or by detecting dipolar splitting with 1 H-13 C NMR. 35,67–69 The measurements are based on different physical interactions, and also the connections between order parameter and quadrupolar or dipolar splitting are different. The absolute values of order parameters from the measured quadrupolar splitting ∆νQ (2 H NMR) are calculated using the equation |SCD | = 4 h ∆νQ , 3 e2 qQ

where the value for the static quadrupole splitting constant is estimated from

various experiments to be 170 kHz leading to a numerical relation |SCD | = 0.00784 × ∆νQ . 66 The absolute values of order parameters from the effective dipolar coupling dCH (1 H-13 C 4πhr 3 i

NMR) are calculated using the equation |SCH | = 2π ¯hµ0 γCH dCH , where values between 20.2– h γc 4πhr 3 i

)−1 , depending on the original authors. 35,67–69 The effective 22.7 kHz are used for (2π ¯hµ0 γCH h γc dipolar coupling dCH is related to the measured dipolar splitting ∆νCH through a scaling factor that depends on the pulse sequence used in the 1 H-13 C NMR experiment. 35,67–69 It is important to note that the order parameters measured with different techniques based on different physical interactions are in good agreement with each other (see Results and Discussion), indicating very high quantitative accuracy of the measurements. For a more detailed discussion, see Ref. 70. The absolute values of order parameters are accessible with both 2 H NMR and 1 H-13 C NMR techniques. However, only 1 H-13 C NMR techniques allow also the measurement of the sign of the order parameter. 16,67,68 The measured sign is negative for almost all the carbons discussed in this work, except for α which is positive. 16,67,68 For more detailed discussion about the determination of the sign of the order parameters, see Ref. 71. For most CH2 segments in a fluid phospholipid bilayer, the order parameters of both hydrogens are equal. However, in some cases (e.g., g1 , g3 , and C2 carbon in the sn-2 chain) the two order parameters are not equal; this can be observed with both 2 H NMR and 1 H-13 C NMR techniques. In the present work, to avoid confusion with the dipolar and quadrupolar

8

ACS Paragon Plus Environment

Page 8 of 48

Page 9 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

splittings in NMR terminology, we call the phenomenon of unequal order parameters for hydrogens attached to the same carbon forking. Forking has been studied in detail with 2

H NMR techniques by deuterating the R or S position in CH2 segment, and the studies

show that it arises from differently sampled orientations of the two C–H bonds, not from two separate populations of lipid conformations. 26,72

Order parameters from simulations The order parameters from simulations were calculated directly using the definition of Eq. 1. For the united atom models, the hydrogen positions were generated post-simulationally from the positions of the heavy atoms and the known hydrocarbon geometries. The statistical error was estimated based on the assumption that different lipids are statistically independent entities (which should be the case in fluid phase): The time average of a given order parameter was first calculated separately for each lipid, and the standard error of the mean over these time averages then taken as the error bar for this order parameter. It has been pointed out that the sampling of individual dihedral angles might be very slow compared to the typical (100 ns) simulation timescales. 73 After 200 ns, however, even 2 ) in the slowest rotational correlation function of a C–H bond (g1 ) reaches a plateau (SCH

the Berger-POPC-07 model 74 —and, notably, the dynamics of this segment have been shown to be significantly slower in simulations than in experiments. 75 In practise, due to averaging over different lipid molecules, less than 200 ns of simulation data should be enough for the order parameter calculation; if the sampling within typical simulation times is not enough for the convergence of the order parameters, then the simulation model in question has unphysically slow dynamics.

Simulated systems All simulations are ran with a standard setup for a planar lipid bilayer in zero tension and constant temperature with periodic boundary conditions in all directions by using the 9

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

GROMACS software package 76 (version numbers 4.5.X–4.6.X), LAMMPS, 77 MDynaMix 78 or NAMD. 79 The number of molecules, simulation temperatures and the length of simulations of all the simulated systems are listed in Tables 1, 2 and 3. Full simulation details are given in the Supporting Information (SI) or in the original publications in case the data is used previously therein. All simulation parameters were set as close to the original parametrization works as possible. Additionally, the files related to the simulations and the resulting trajectories are publicly available for almost all systems in the Zenodo collection https://zenodo.org/collection/user-nmrlipids. The references pointing to simulation details and files are also listed in Tables 1, 2 and 3.

Results and Discussion Full hydration: Experimental order parameters for the glycerol backbone and headgroup The specific deuteration of α-, β- and g3 - segments of DPPC has been successful, allowing the absolute values of the order parameters for these segments to be measured by 2 H NMR. 48–50,54 In addition, the absolute values of order parameters for all glycerol backbone and choline headgroup segments in egg yolk lecithin, 67 DMPC, 16,68,69 DOPC, 141 and POPC 35,141 have been measured with several different implementations of 1 H-13 C NMR experiments. Furthermore, for some systems the signs of the order parameters have been measured with 1 H-13 C NMR techniques. 16,67,68 The experimental values of the glycerol backbone and choline order parameters from various publications, 35,50,54,68,69 with the signs measured in Refs. 16,67,68, are shown in Fig. 2. In general there is a good agreement between the order parameters measured with different experimental NMR techniques: Almost all the reported values are within a variation of ±0.02 (which is also the error estimate given by Gross et al. 68 ) for all fully hydrated PC bilayers, regardless of variation in their acyl chain composition and temperature. Exceptions 10

ACS Paragon Plus Environment

Page 10 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Force field Berger-DMPC-04 80 Berger-DPPC-98 83 Berger-POPC-07 74 CHARMM36 31 CHARMM36 31 CHARMM36 31 CHARMM36 31 MacRog 89 MacRog 89 MacRog 89 GAFFlipid 33 GAFFlipid 33 GAFFlipid 33 Lipid14 95 Poger 97 Slipids 100 Slipids 102 Kukol 104 Chiu 106 H¨ogberg08 29 H¨ogberg08 109 Ulmschneiders 111 Tj¨ornhammar14 113 Botan-CHARMM36-UA 115 Lee-CHARMM36-UA 117

lipid DMPC DPPC POPC DPPC DPPC POPC POPC POPC POPC POPC DPPC DPPC POPC POPC DPPC DPPC POPC POPC POPC DMPC POPC POPC DPPC DLPC DPPC

a

b Nw Nl 128 5097 72 2864 128 7290 72 2189 72 2189 72 2242 128 5120 288 12600 128 6400 288 14400 72 2197 72 2167 126 3948 72 2234 128 5841 128 3840 128 5120 512 20564 128 3552 98 3840 128 3840 128 3328 144 7056 128 3840 72 2189

c

T (K) 323 323 298 323 323 303 303 310 310 310 323 323 303 303 323 323 303 298 298 303 303 310 323 323 323

d

tsim (ns) 130 60 270 30 130 30 200 100 400 90 90 250 137 100 2×100 150 200 50 56 75 100 100 200 30 70

e

tanal (ns) 100 30 240 25 20 100 80 200 40 50 250 32 50 2×50 100 150 30 50 50 80 50 100 20 50

Files [ 81]∗ [ 84]∗ [ 85]∗ [ 86]∗ [ 87]∗ [ 88]∗ [ 90]∗ [ 91]∗ [ 92]∗ [ 93]∗ [ 94]∗ [ 96]∗ [ 98,99]∗ [ 101]∗ [ 103]∗ [ 105]∗ [ 107]∗ [ 108]∗ [ 110]∗ [ 112]∗ [ 114]∗ [ 116] [ 118]∗

f

g

Details [ 82] SI [ 75] SI [ 31]h SI SI SI SI SI SI [ 33]i SI SI SI SI SI SI SI [ 29] [ 109] SI [ 113] SI SI

Table 1: Fully hydrated single component lipid bilayer systems simulated for Fig. 2: 1,2-dimyristoyl-sn-glycero3-phosphocholine (DMPC), dilauroylphosphatidylcholine (DLPC), dipalmitoylphosphatidylcholine (DPPC), and 1-palmitoyl-2-oleoylphosphatidylcholine (POPC). The bolded systems were used also for Fig. 3. a Number of lipid molecules. b Number of water molecules. c Temperature. d Total simulation time. e Time used for analysis. f Reference link for the downloadable simulation files; the data sets marked with ∗ also include a part of the trajectory. g Reference for the full simulation details; the original publication is cited if simulation data from previously published work has been directly used, for other systems the simulation details are given in the Supporting Information. h Magnitudes from Fig. S4 of Klauda et al., 31 signs matched to our simulations. i Magnitudes from Fig. 9 of Dickson et al., 33 signs matched to our simulations.

Page 11 of 48 The Journal of Physical Chemistry

ACS Paragon Plus Environment

11

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 12 of 48

Page 13 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Table 2: Simulated single component lipid bilayers with varying hydration levels. The simulation file data sets marked with ∗ include also part of the trajectory. a Water/lipid molar ratio b The number of lipid molecules c The number of water molecules d Simulation temperature e The total simulation time f Time frames used in the analysis g Reference link for the downloadable simulation files h Reference for the full simulation details Force field Berger-POPC-07 74 Berger-DLPC-13 120

CHARMM36 31

MacRog 89

GAFFlipid 33

lipid POPC POPC DLPC DLPC DLPC DLPC DLPC DLPC DLPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC

a

n (w/l) 57 7 28 24 20 16 12 8 4 40 31 15 7 50 44 25 20 15 10 5 31 7

b

Nl 128 128 72 72 72 72 72 72 72 128 72 72 72 288 288 288 288 288 288 288 126 126

c

Nw 7290 896 2016 1728 1440 1152 864 576 288 5120 2242 1080 504 14400 12600 7200 5760 4320 2880 1440 3948 896

d

T (K) 298 298 300 300 300 300 300 300 300 303 303 303 303 310 310 310 310 310 310 310 303 303

e

tsim (ns) 270 60 80 80 80 80 80 80 80 150 30 59 60 90 100 100 100 100 100 100 137 130

f

tanal (ns) 240 50 60 60 60 60 60 60 60 100 20 40 20 40 80 50 50 50 50 50 32 40

g

Files [ 85]∗ [ 119]∗ [ 121]∗ [ 122]∗ [ 123]∗ [ 124]∗ [ 125]∗ [ 126]∗ [ 127]∗ [ 88]∗ [ 87]∗ [ 128]∗ [ 129]∗ [ 92]∗ [ 90]∗ [ 92]∗ [ 92]∗ [ 92]∗ [ 92]∗ [ 92]∗ [ 94]∗ [ 130]∗

are the somewhat lower order parameters reported from some measurements using 1 H-13 C NMR. 16,67,141 In these experiments, however, the reported error bars are either relatively large, 16,67 or the spectral resolution is quite low and numerical lineshape simulations have not been used in the analysis. 141 As it, therefore, is highly likely that the reported lower order parameters are due to lower experimental accuracy, we exclude them from the present discussion; for more details, see Ref. 70. Motivated by the high experimental reproducibility, we have highlighted in Fig. 2 subjective sweet spots (light blue areas spanning 0.04 units around the average of the extremal experimental values), within which we expect the

13

ACS Paragon Plus Environment

h

Details SI SI [ 120] [ 120] [ 120] [ 120] [ 120] [ 120] [ 120] SI SI SI SI SI SI SI SI SI SI SI SI SI

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Paragon Plus Environment

14

MacRog 89

CHARMM36 31,137

Force field Berger-POPC-07 74 /H¨oltje-CHOL-13 35,131

a

Nl 128 120 110 84 64 50 128 512 460 436 100 410 384 332 256 80 128 114 72 64 56

lipid POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC POPC

b

8 18 44 64 78 0 0 52 76 24 102 128 180 256 80 0 14 56 64 72

Nchol 0

c

6% 14% 34% 50% 61% 0% 0% 10% 15% 19% 20% 25% 35% 50% 50% 0% 11% 44% 50% 56%

CCHOL 0% 7290 8481 6794 10314 5782 5120 23943 23569 23331 4960 20972 22327 21340 20334 4496 6400 6400 6400 6400 6400

Nw 7290

d

e

298 298 298 298 298 303 298 298 298 303 298 298 298 298 303 310 310 310 310 310

T (K) 298

f

100 100 100 100 100 150 170 170 170 200 170 170 170 170 200 400 400 400 400 400

tsim (ns) 270

g

80 80 80 80 80 100 100 100 100 100 100 100 100 100 100 200 200 200 200 200

tanal (ns) 240

[ 132]∗ [ 133]∗ [ 134]∗ [ 135]∗ [ 136]∗ [ 88]∗ [ 138]∗ [ 138]∗ [ 138]∗ [ 139]∗ [ 138]∗ [ 138]∗ [ 138]∗ [ 138]∗ [ 140]∗ [ 91]∗ [ 91]∗ [ 91]∗ [ 91]∗ [ 91]∗

Files [ 85]∗

h

i

[ [ [ [ [

35] 35] 35] 35] 35] SI SI SI SI SI SI SI SI SI SI SI SI SI SI SI

Details [ 75]

Table 3: Simulated lipid bilayers containing cholesterol. The simulation file data sets marked with ∗ include also part of the trajectory. a The number of lipid molecules b The number of cholesterol molecules c Cholesterol concentration (mol%) d The number of water molecules e Simulation temperature f The total simulation time g Time frames used in the analysis h Reference link for the downloadable simulation files i Reference for the full simulation details

The Journal of Physical Chemistry Page 14 of 48

Page 15 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

calculated values of the order parameters of a well-performing force field to fall. In addition to the numerical values, forking (see Sec. ) is an important feature of the order parameters. In contrast to the lack of forking in the choline segments α and β, both CH2 segments of the glycerol backbone fork. In the g3 segment forking is small (≈ 0.02), and some experiments only report the larger or the average value. 35,50 However, forking is significant for the g1 segment, whose lower order parameter is close to zero and the larger one has an absolute value of approximately 0.13–0.15. Forking was studied in detail by Gally et al., 26 who used E. Coli to stereospecifically deuterate the different hydrogens attached to the g1 or g3 groups in PE lipids, and measured the order parameters from the lipid extract. This experiment gave the lower order parameter when deuterium was in the S position of g1 or R position for g3 . Since the glycerol backbone order parameters are very similar irrespective of the headgroup chemistry (PC, PE, or PG) or lipid environment, 26 it is reasonable to assume that the stereospecifity measured for the PE lipids holds also for the PC lipids. The most detailed experimentally available order parameter information for the glycerol backbone and choline segments of POPC bilayer is collected by taking the absolute values from Ref. 35, the signs from Refs. 16,67,68 and the stereospecific labeling from Ref. 26, and is shown in Fig. 3.

Full hydration: Comparison between simulation models and experiments The order parameters of the glycerol backbone and headgroup calculated from different force fields for various lipids have been previously compared to experiments. 28–37 The general conclusion from these studies seems to be that the CHARMM based, 29,31 GAFFlipid 33 and MacRog 37 force fields perform better for the glycerol backbone and headgroup structures than the GROMOS based models. 30,32,34,35 However, none of the studies exploits the full potential of the available experimental data discussed in the previous section, i.e. the quantitative accuracy, known signs and stereospecific labeling of the experimental order 15

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 16 of 48

Page 17 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

parameters. To get a general idea of the quality of the glycerol backbone and choline headgroup structures in different models, we calculated the order parameters for these parts from thirteen different lipid models (Table 1) and plotted the results together with experimental values in Fig. 2. Two criteria were used to judge the quality of the model: 1) there must not be significant forking in the α and β carbons, there must be only moderate forking in the g3 carbon and there must be significant forking in the g1 carbon, and 2) the magnitude should be preferably inside to the subjective sweet spots determined from experiments (blue shaded regions in Fig. 2). The results for each force field with respect to the above criteria are summarized in Figure 4. None of the studied force fields fulfils these criteria completely, however CHARMM36 is close. This is not surprising since the dihedral potentials in this model are tuned against experiments to better reproduce these order parameters. 31 The next models in the list are CHARMM36-UA 115,117 and H¨ogberg08, 29 which is also not surprising since these models are using the CHARMM bonded potentials for glycerol backbone and choline. The fourth and the fifth models in the list, MacRog 37 and GAFFlipid, 33 have independently determined dihedral potentials. All the models based on Gromos potentials and Slipids perform less well. In the following sections we subject CHARMM36, MacRog, GAFFlipid and Berger-POPC07 to a more careful comparison including the stereospecific labeling (Fig. 3), atomistic level structure, and responses to dehydration and cholesterol content. These models are selected for more detailed study since they are the best representatives of different dihedral potential parametrization techniques (CHARMM36, MacRog, GAFFlipid), and the Berger based models are the most used lipid models in the literature.

Full hydration: Atomistic resolution structures in different models The results in the previous section revealed significant differences of the glycerol backbone and choline headgroup order parameters between different molecular dynamics simulation 17

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 18 of 48

Page 19 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

models. However, it is not straightforward to conclude which kind of structural differences (if any) between the models the results indicate, because the mapping from the order parameters to the structure is not unique. In this section we demonstrate that 1) the differences in order parameters indicate significantly different structural sampling, which is strongly correlated with the dihedral angles of the related bonds; and that 2) the comparison between experimental and simulated order parameters can be used to exclude nonrealistic structural sampling in molecular dynamics simulations. The demonstration is done for the dihedral angles defined by the g3 -g2 -g1 -O(sn-1) segments in the glycerol backbone and the N-β-α-O segments in the headgroup. These dihedrals were chosen for demonstration, because significant differences between the models are observed around these segments in Fig. 3. We note that performing a similar comparison through all the dihedrals in all the 13 models would probably give highly useful information on how to improve the accuracy of the models; yet this is beyond the scope of the current report. The dihedral angle distributions for the g3 -g2 -g1 -O(sn-1) dihedral calculated from different models are shown in Fig. 5. The distribution is qualitatively different for the BergerPOPC-07 model, showing a maximum in the gauche+ -conformation (60o ) compared to all the other models showing a maximum in the anti-conformation (180o ). The distributions in all the other models have the same general features, the main difference being that the fraction of configurations in the gauche− -conformation (-60o ) is zero for the MacRog, detectable for the CHARMM36 and equally large to the gauche+ fraction in GAFFlipid. From the results we conclude that most likely the wrongly sampled dihedral angle for the g2 -g1 bond explains the significant discrepancy to the experimental order parameters for the g1 segment in the Berger-POPC-07 model (Fig. 3). In conclusion, models preferring the anti conformation for this dihedral give more realistic order parameters; this is in agreement with previous crystal structure and 1 H NMR studies. 19–21,23–25 The dihedral angle distribution for the N-β-α-O dihedral calculated from the same four models is shown in Fig. 6. Also for this dihedral there are significant differences in the

19

ACS Paragon Plus Environment

Berger-POPC-07

g 3 -g 2 -g 1 -O(sn-1)

-180 -120

-60

0

60

120

-180 -120

-60

0

60

Page 20 of 48

CHARMM36

180

-180 -120

-60

0

60

120

180

120

180

GAFFLipid

MacRog

120

180

Dihedral angle (degrees)

-180 -120

-60

0

60

Dihedral angle (degrees)

Figure 5: Dihedral angle distributions for g3 -g2 -g1 -O(sn-1) dihedral from different models (POPC bilayer in full hydration). gauche–anti fractions. The gauche conformations are dominant in CHARMM36, in MacRog there are only anti conformations present, and in Berger-POPC-07 and GAFFlipid the gauche and anti conformations have equal probabilities. On the other hand, comparison of α and β order parameters in Fig. 3 reveals that for these carbons the CHARMM36 is closest to the experimental results and it is also the only model that has the correct sign (negative) for the β order parameter. This result is again in agreement with previous crystal structure, 1

H NMR, and Raman spectroscopy studies, 19–22 which suggest that this dihedral is in the

N- β -α -O

gauche conformation in the absence of ions. Berger-POPC-07

-180 -120

-60

0

60

120

CHARMM36

180

-180 -120

-180 -120

-60

0

60

-60

0

60

120

180

120

180

GAFFLipid

MacRog

N- β - α-O

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

g 3 -g2 -g 1 -O(sn-1)

The Journal of Physical Chemistry

120

180

Dihedral angle (degrees)

-180 -120

-60

0

60

Dihedral angle (degrees)

Figure 6: Dihedral angle distributions for N-β-α-O dihedral from different models (POPC bilayer in full hydration). 20

ACS Paragon Plus Environment

Page 21 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

These examples show that the glycerol backbone and headgroup order parameters reflect the atomistic resolution structure and that the comparison with experiments allows the assessment of the quality of the suggested structure. We were able to pinpoint specific problems in the structures in different models and suggest potential improvement strategies. If an improved atomistic molecular dynamics simulation model would reproduce the order parameters and other experimental observables (like 31

31

P chemical shift anisotropy 36 and

P-13 C dipolar couplings 43 ) within experimental accuracy, it would give an interpretation

for the atomistic resolution structure of the glycerol backbone and choline. 10–13,15,16,18 The research along these lines is left, however, for future studies.

Response to dehydration and cholesterol content In addition to pure phosphatidylcholine bilayers at full hydration, the choline headgroup order parameters have been measured under various different conditions. 30,32,35,45–51,54,55 Also the order parameters for the glycerol backbone have been measured with 1 H-13 C NMR in dehydrated conditions, 47 and as a function of anesthetics 30 and glycolipids 32 for DMPC, and as a function of cholesterol concentration for POPC. 35 Due to the high resolution in the NMR (especially 2 H NMR) experiments, even very small order parameter changes resulting from the varying conditions can be measured (see Ref. 70 for more discussion)—but, as already discussed above, it is not simple to deduce the structural changes from order parameter changes. 15,18 However, comparison of the order parameters between simulations and experiments in different conditions can be used to assess the quality of the force field in different situations, and, if the quality is good, to interpret the structural changes in experiments. Here we exemplify such comparison for a lipid bilayer under low hydration levels and when varying amounts of cholesterol is included in the bilayer. The interaction between ions and a phosphatidylcholine bilayer will be discussed in a separate study. 60

21

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 48

Phospholipid bilayer with low hydration level Fig. 7 shows the published 45–47 experimental order parameters for the glycerol backbone and choline as a function of hydration level. Despite slight differences in temperature and acyl chain composition, the three independently reported data sets for the choline (β and α) segments agree well with each other: Both order parameters increase with decreasing hydration level. The glycerol backbone order parameters (g3 , g2 , g1 ), in contrast, have been observed 47 to slightly decrease with dehydration. Note that the original experiments 45–47 measured only absolute values, but here we included the signs measured in separate studies. 16,67,68 Consequently, the negative β order parameter actually increases with dehydration as its absolute value decreases. 45–47 Lipid bilayer dehydration has been studied also with molecular dynamics simulations, 142–147 typically motivated by the discussion concerning the origin of the “hydration repulsion”. 148–150 Only one 142 of these studies, however, compared their simulation model to the experimental choline and glycerol backbone order parameters. Fig. 7 shows these order parameters as a function of hydration level for the CHARMM36, MacRog and GAFFlipid models (having the most realistic atomistic resolution structures) and a Berger-based model (which is the most used lipid model); note that the simulation results have been vertically shifted to ease the comparison with experimental response to dehydration. Despite of some fluctuations, the increase of the choline (β and α) order parameters is seen in all four models. The response of the choline order parameters to dehydration can, therefore, be interpreted to qualitatively agree with experiments. The situation is significantly more complicated for the glycerol backbone: None of the four models produced the experimentally seen trends in all the (g3 , g2 , g1 ) segments. The qualitative agreement of the α and β order parameters with experiments in all four simulation models (Fig. 7) indicates that, despite the unrealistic structures at full hydration (Figs. 2 and 4), the structural response of the choline headgroup to dehydration is somewhat realistic. A likely explanation is that as the interlamellar space shrinks with dehydration, 22

ACS Paragon Plus Environment

Page 23 of 48

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 24 of 48

Page 25 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Cholesterol-containing phospholipid bilayer As cholesterol is abundant in biological membranes and has been suggested to be an important player, for example, in domain formation, 151,152 phospholipid–cholesterol interactions have been extensively studied with theoretical 153–156 and experimental 8,35,48,157 methods. It is widely agreed that cholesterol orders lipid acyl tails and thus decreases the area per molecule (condensing effect), but its influence on the lipid headgroup and glycerol backbone remains debated. 151–153 It has been suggested, for example, that the surrounding phospholipids shield cholesterol from exposure to water by reorienting their headgroups (“umbrella model”) 153 or that cholesterol acts as a spacer between the headgroups to increase their entropy and dynamics (“superlattice model”). 152 Molecular dynamics simulations have supported both the umbrella 156 as well as the superlattice 154 model, in addition to suggesting specific interactions of cholesterol with the glycerol backbone. 155 In these studies 154–156 the responses of the glycerol backbone and choline headgroup to increasing cholesterol content were not, however, compared to experiments. Fig. 9 shows the responses of the choline headgroup (β and α) order parameters of POPC (measured by 1 H-13 C NMR 35 ) and DPPC (2 H NMR 48 ) to increasing cholesterol content. Again, the two independent data sets agree very well: Only very modest (∆SCH < 0.03) changes occur in the choline order parameters as cholesterol content increases from 0 to 60%. The extreme sensitivity of the high resolution 2 H NMR experiments is beautifully demonstrated by the measurable 48 (but barely visible on the scale used in Fig. 9) cholesterolinduced forking of the α order parameter. We note that the modest (∆SCH < 0.02 for g1 ; < 0.04 for g2 , g3 ; see Fig. 9) effects of cholesterol on the glycerol backbone order parameters of POPC measured by 1 H-13 C NMR 35 agree well with the results for phosphatidylethanolamine (PE) measured by 2 H NMR. 158 This further supports the ideas that the glycerol backbone structural behaviour is independent of the headgroup composition 26 and that the headgroup structure is largely independent of the acyl chain region content unless charges are present. 27 25

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 26 of 48

Page 27 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

In addition to the experimental data, Fig. 9 shows our results for the CHARMM36 and MacRog force fields and the previously published 35 Berger-POPC-07/H¨oltje-CHOL-13 results. Note that the simulation data are shifted vertically to ease comparison with experimental responses. As previously pointed out, 35 the Berger-based model seriously exaggerates the effect of cholesterol on the phospholipid glycerol backbone and choline headgroup. In comparison, the choline and glycerol backbone responses of CHARMM36 and MacRog are in better qualitative agreement with experiments. Therefore, to resolve the nature of cholesterol-induced structural changes, we calculated from CHARMM36 the glycerol backbone orientation and dihedral angle distributions at various cholesterol contents (Supporting Information). The only detectable changes are the small decrease of gauche(-) and increase of gaughe(+) probability of the g3 -g2 -g1 -O(sn-1) dihedral and slight (less than 5 degrees) change in the glycerol backbone orientation. In conclusion, our results suggest that the significant effects of cholesterol on lipid conformations observed in simulations 154–156 are overestimated by the computational models used; cholesterol only induces very small structural changes in the glycerol backbone. Finally, it is important to note that the CHARMM36 force field parameters (glycerol backbone dihedral potentials) have been tuned to reproduce the experimental order parameters at full hydration. 31 This approach introduces a risk of overfitting, which would manifest itself as wrong responses to changing conditions. Interestingly, according to our results, tuning did not lead to overfitting problems as far as dehydration or cholesterol content are considered.

27

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Conclusions The atomistic resolution structures sampled by the glycerol backbone and choline headgroup in phoshatidylcholine bilayers are not known despite of vast amount of accurate experimental data. An atomistic resolution molecular dynamics simulation model that would reproduce the experimental data would automatically resolve the structures, thus giving an unprecedently detailed interpretation of the experimental data. In this work we have collected and reviewed the experimental C–H bond vector order parameters available in the literature. These accurate experimental data are then compared to 13 different atomistic resolution simulation models for a fully hydrated lipid bilayer system, followed by bilayers dehydrated to different extents, and finally bilayers containing various amounts of cholesterol. We are led to the following four main conclusions: (1) The C–H bond order parameters measured with different NMR techniques are consistent. By combining the experimental results from various sources we concluded that the order parameters for each C–H bond are known with a quantitative accuracy of ±0.02. (2) Comparison of order parameters between experiments and different atomistic resolution models together with structural analysis showed that the order parameters can be used to judge the structural accuracy of a model. Thus the combination of atomistic resolution molecular dynamics simulations and NMR experiments can be used to resolve the atomistic resolution structures of biomolecules in biologically relevant conditions. This approach can be extended from lipids to, for example, membrane proteins. (3) The review of previous experimental results revealed that when a bilayer is dehydrated the choline order parameters increase. Our simulations suggested that this can be explained by the P–N vector tilting more parallel to the membrane. This strongly supports and complements the idea that charge-induced choline tilting can be measured using order parameter changes. 55,60 (4) Only modest changes of glycerol backbone and choline order parameters are observed experimentally with increasing cholesterol content. When interpreted using the computa28

ACS Paragon Plus Environment

Page 28 of 48

Page 29 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

tional lipid model that we found to have the most realistic response to cholesterol, this observation means that cholesterol induces only minor changes in the g3 -g2 -g1 -O(sn-1) dihedral of the glycerol backbone, in other words, there is no major conformational change of the lipid. (+) Besides these four main conclusions, we note that we have created the most extensive publicly available collection of molecular dynamics simulation trajectories of lipid bilayers (https://zenodo.org/collection/user-nmrlipids). The mere existence of this collection opens up numerous possibilities for unforeseen analyses, such as data mining, and rapid testing of ideas with much less computational effort than previously required. In general, we conclude that in order to fully utilize the potential of atomistic-resolution classical molecular dynamics simulations in the structural interpretation of high resolution NMR data 159 for lipid bilayers, one must improve the phoshatidylcholine glycerol backbone and choline headgroup parameters of the existing force fields. This work has been done as a fully open collaboration, using nmrlipids.blogspot.fi as the communication platform. All the scientific contributions have been communicated publicly through this blog. 61

Acknowledgement We acknowledge all the discussion participants at nmrlipids.blogspot.fi. AB and CL acknowledge financial support from the French National Research Agency (ANR: Biolubrication by phospholipid membranes, Biolub2012) and computing time allocation from Pˆole Scientifique de Mod´elisation Num´erique from the ENS Lyon (PSMN), and Centre Informatique National de l’Enseignement Sup´erieur (CINES, Montpelier, France) (Project c2015096850). FFR acknowledges CONACYT and DGAPA UNAM IG100513 for financial support, Cluster H´ıbrido de Superc´omputo Xiuhcoatl - CINVESTAV and Miztli - UNAM for computational resources. MJ and JT acknowledge and CSC – IT Center for Science for computational resources (project number tty3979). MJ also acknowledges the Finnish Doctoral Programme in 29

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Computational Sciences (FICS) for funding. JT, MJ, TR and WK acknowledge the funding from the Academy of Finland (Centre of Excellence program) and the European Research Council (Advanced Grant project CROWDED-PRO-LIPIDS). WK acknowledges CSC – IT Centre for Science (Espoo, Finland) for excellent computational resources (project number tty3995). MSM acknowledges financial support from the Volkswagen Foundation (86110). LM acknowledges funding proved by the Institut national de la sant´e et de la recherche m´edicale (INSERM). JM acknowledges CSC – IT Center for Science for computational resources. OHSO acknowledges Tiago Ferreira and Paavo Kinnunen for useful discussions, the Emil Aaltonen foundation for financial support, Aalto Science – IT project and CSC – IT Center for Science for computational resources. HS acknowleges Catherine Etchebest and St´ephane T´eletch´ea for useful discussions and continued support, the HPC resources granted from GENCI-CINES (Grant 2014-c2014077209) and computer facilities provided by R´egion Ile de France and INTS (SESAME 2009 project).

Supporting Information Available Simulation and analysis details, two figures, and author contributions. This material is available free of charge via the Internet at http://pubs.acs.org/.

References (1) Lipowsky, R., Sackmann, E., Eds. Structure and Dynamics of Membranes; Elsevier, 1995. (2) Tieleman, D. P.; Marrink, S. J.; Berendsen, H. J. C. A computer perspective of membranes: molecular dynamics studies of lipid bilayer systems. Biochim. Biophys. Acta 1997, 1331, 235–270. (3) Klauda, J. B.; Venable, R. M.; Jr., A. D. M.; Pastor, R. W. In Computational Modeling

30

ACS Paragon Plus Environment

Page 30 of 48

Page 31 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

of Membrane Bilayers; Feller, S. E., Ed.; Current Topics in Membranes; Academic Press, 2008; Vol. 60; pp 1 – 48. (4) Edholm, O. In Computational Modeling of Membrane Bilayers; Feller, S. E., Ed.; Current Topics in Membranes; Academic Press, 2008; Vol. 60; pp 91 – 110. (5) Tieleman, D. P. In Molecular Simulations and Biomembranes: From Biophysics to Function; Sansom, M., Biggin, P., Eds.; The Royal Society of Chemistry, 2010; pp 1–25. ´ Khalid, S. Molecular Dynamics Simulations of Phosphatidyl(6) Piggot, T. J.; Pi˜ neiro, A.; choline Membranes: A Comparative Force Field Study. J. Chem. Theory Comput. 2012, 8, 4593–4609. (7) Rabinovich, A.; Lyubartsev, A. Computer simulation of lipid membranes: Methodology and achievements. Polymer Science Series C 2013, 55, 162–180. (8) Marsh, D. Handbook of Lipid Bilayers, Second Edition; RSC press, 2013. (9) Israelachvili, J. N.; Marcelja, S.; Horn, R. G. Physical Principles of Membrane Organization. Q. Rev. Biophys. 1980, 13, 121–200. (10) Seelig, J.; Gally, H.-U.; Wohlgemuth, R. Orientation and flexibility of the choline head group in phosphatidylcholine bilayers. Biochim. Biophys. Acta 1977, 467, 109 – 119. (11) Skarjune, R.; Oldfield, E. Physical studies of cell surface and cell membrane structure. Determination of phospholipid head group organization by deuterium and phosphorus nuclear magnetic resonance spectroscopy. Biochemistry 1979, 18, 5903–5909. (12) Jacobs, R. E.; Oldfield, E. {NMR} of membranes. Prog. Nucl. Mag. Res. Sp. 1980, 14, 113 – 136. (13) Davis, J. H. The description of membrane lipid conformation, order and dynamics by 2H-NMR. Biochim. Biophys. Acta 1983, 737, 117 – 171. 31

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(14) Strenk, L.; Westerman, P.; Doane, J. A model of orientational ordering in phosphatidylcholine bilayers based on conformational analysis of the glycerol backbone region. Biophys. J. 1985, 48, 765 – 773. (15) Akutsu, H.; Nagamori, T. Conformational analysis of the polar head group in phosphatidylcholine bilayers: a structural change induced by cations. Biochemistry 1991, 30, 4510–4516. (16) Hong, M.; Schmidt-Rohr, K.; Nanz, D. Study of phospholipid structure by 1H, 13C, and 31P dipolar couplings from two-dimensional {NMR}. Biophys. J. 1995, 69, 1939 – 1950. (17) Hong, M.; Schmidt-Rohr, K.; Zimmermann, H. Conformational Constraints on the Headgroup and sn-2 Chain of Bilayer DMPC from NMR Dipolar Couplings. Biochemistry 1996, 35, 8335–8341. (18) Semchyschyn, D. J.; Macdonald, P. M. Conformational response of the phosphatidylcholine headgroup to bilayer surface charge: torsion angle constraints from dipolar and quadrupolar couplings in bicelles. Magn. Res. Chem. 2004, 42, 89–104. (19) Hauser, H.; Guyer, W.; Pascher, I.; Skrabal, P.; Sundell, S. Polar group conformation of phosphatidylcholine. Effect of solvent and aggregation. Biochemistry 1980, 19, 366–373. (20) Hauser, H.;

Guyer, W.;

Paltauf, F. Polar group conformation of 1,2-di-O-

alkylglycerophosphocholines in the absence and presence of ions. Chem. Phys. Lipids 1981, 29, 103 – 120. (21) Hauser, H.; Pascher, I.; Pearson, R.; Sundell, S. Preferred conformation and molecular packing of phosphatidylethanolamine and phosphatidylcholine. Biochim. Biophys. Acta 1981, 650, 21 – 51.

32

ACS Paragon Plus Environment

Page 32 of 48

Page 33 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(22) Akutsu, H. Direct determination by Raman scattering of the conformation of the choline group in phospholipid bilayers. Biochemistry 1981, 20, 7359–7366. (23) Pascher, I.; Lundmark, M.; Nyholm, P.-G.; Sundell, S. Crystal structures of membrane lipids. Biochim. Biophys. Acta 1992, 1113, 339 – 373. (24) Hauser, H.; Pascher, I.; Sundell, S. Preferred conformation and dynamics of the glycerol backbone in phospholipids. An NMR and x-ray single-crystal analysis. Biochemistry 1988, 27, 9166–9174. (25) Marsh, D.; P´ali, T. Lipid conformation in crystalline bilayers and in crystals of transmembrane proteins. Chem. Phys. Lipids 2006, 141, 48 – 65. (26) Gally, H. U.; Pluschke, G.; Overath, P.; Seelig, J. Structure of Escherichia coli membranes. Glycerol auxotrophs as a tool for the analysis of the phospholipid head-group region by deuterium magnetic resonance. Biochemistry 1981, 20, 1826–1831. (27) Scherer, P.; Seelig, J. Structure and dynamics of the phosphatidylcholine and the phosphatidylethanolamine head group in L-M fibroblasts as studied by deuterium nuclear magnetic resonance. The EMBO journal 1987, 6 . (28) Shinoda, W.; Namiki, N.; Okazaki, S. Molecular dynamics study of a lipid bilayer: Convergence, structure, and long-time dynamics. J. Chem. Phys. 1997, 106, 5731– 5743. (29) H¨ogberg, C.-J.; Nikitin, A. M.; Lyubartsev, A. P. Modification of the CHARMM force field for DMPC lipid bilayer. J. Comput. Chem. 2008, 29, 2359–2369. (30) Castro, V.; Stevensson, B.; Dvinskikh, S. V.; H¨ogberg, C.-J.; Lyubartsev, A. P.; Zimmermann, H.; Sandstr¨om, D.; Maliniak, A. {NMR} investigations of interactions between anesthetics and lipid bilayers. Biochim. Biophys. Acta - Biomembranes 2008, 1778, 2604 – 2611. 33

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(31) Klauda, J. B.; Venable, R. M.; Freites, J. A.; O’Connor, J. W.; Tobias, D. J.; Mondragon-Ramirez, C.; Vorobyov, I.; Jr, A. D. M.; Pastor, R. W. Update of the CHARMM All-Atom Additive Force Field for Lipids: Validation on Six Lipid Types. J. Phys. Chem. B 2010, 114, 7830–7843. (32) Kapla, J.; Stevensson, B.; Dahlberg, M.; Maliniak, A. Molecular Dynamics Simulations of Membranes Composed of Glycolipids and Phospholipids. J. Phys. Chem. B 2012, 116, 244–252. (33) Dickson, C. J.; Rosso, L.; Betz, R. M.; Walker, R. C.; Gould, I. R. GAFFlipid: a General Amber Force Field for the accurate molecular dynamics simulation of phospholipid. Soft Matter 2012, 8, 9617–9627. (34) Poger, D.; Mark, A. E. Lipid Bilayers: The Effect of Force Field on Ordering and Dynamics. J. Chem. Theory Comput. 2012, 8, 4807–4817. (35) Ferreira, T. M.; Coreta-Gomes, F.; Ollila, O. H. S.; Moreno, M. J.; Vaz, W. L. C.; Topgaard, D. Cholesterol and POPC segmental order parameters in lipid membranes: solid state 1H-13C NMR and MD simulation studies. Phys. Chem. Chem. Phys. 2013, 15, 1976–1989. (36) Chowdhary, J.; Harder, E.; Lopes, P. E. M.; Huang, L.; MacKerell, A. D.; Roux, B. A Polarizable Force Field of Dipalmitoylphosphatidylcholine Based on the Classical Drude Model for Molecular Dynamics Simulations of Lipids. J. Phys. Chem. B 2013, 117, 9142–9160. (37) Maciejewski, A.; Pasenkiewicz-Gierula, M.; Cramariuc, O.; Vattulainen, I.; Rog, T. Refined OPLS All-Atom Force Field for Saturated Phosphatidylcholine Bilayers at Full Hydration. J. Phys. Chem. B 2014, 118, 4571–4581. (38) Robinson, A.; Richards, W.; Thomas, P.; Hann, M. Head group and chain behavior in

34

ACS Paragon Plus Environment

Page 34 of 48

Page 35 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

biological membranes: a molecular dynamics computer simulation. Biophys. J. 1994, 67, 2345 – 2354. (39) Essex, J. W.; Hann, M. M.; Richards, W. G. Molecular Dynamics Simulation of a Hydrated Phospholipid Bilayer. Philos. T. Roy. Soc. B 1994, 344, 239–260. (40) Kothekar, V. Molecular dynamics simulation of hydrated phospholipid bilayers. Ind. J. Biochem. Biophys. 1996, 33, 431 – 447. (41) Hyv¨onen, M. T.; Rantala, T. T.; Ala-Korpela, M. Structure and Dynamic Properties of Diunsaturated 1-Palmitoyl-2-Linoleoyl-sn-Glycero-3-Phosphatidylcholine Lipid Bilayer from Molecular Dynamics Simulation. Biophys. J. 1997, 73, 2907–2923. (42) Duong, T. H.; Mehler, E. L.; Weinstein, H. Molecular Dynamics Simulation of Membranes and a Transmembrane Helix. J. Comput. Phys. 1999, 151, 358 – 387. (43) Prakash, P.; Sankararamakrishnan, R. Force field dependence of phospholipid headgroup and acyl chain properties: Comparative molecular dynamics simulations of DMPC bilayers. J. Comp. Chem. 2010, 31, 266–277. (44) Berger, O.; Edholm, O.; J¨ahnig, F. Molecular dynamics simulations of a fluid bilayer of dipalmitoylphosphatidylcholine at full hydration, constant pressure, and constant temperature. Biophys. J. 1997, 72, 2002 – 2013. (45) Bechinger, B.; Seelig, J. Conformational changes of the phosphatidylcholine headgroup due to membrane dehydration. A 2H-NMR study. Chem. Phys. Lipids 1991, 58, 1 – 5. (46) Ulrich, A.; Watts, A. Molecular response of the lipid headgroup to bilayer hydration monitored by 2H-NMR. Biophys. J. 1994, 66, 1441 – 1449. (47) Dvinskikh, S. V.; Castro, V.; Sandstrom, D. Probing segmental order in lipid bilayers

35

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

at variable hydration levels by amplitude- and phase-modulated cross-polarization NMR. Phys. Chem. Chem. Phys. 2005, 7, 3255–3257. (48) Brown, M. F.; Seelig, J. Influence of cholesterol on the polar region of phosphatidylcholine and phosphatidylethanolamine bilayers. Biochemistry 1978, 17, 381–384. (49) Brown, M. F.; Seelig, J. Ion-induced changes in head group conformation of lecithin bilayers. Nature 1977, 269, 721–723. (50) Akutsu, H.; Seelig, J. Interaction of metal ions with phosphatidylcholine bilayer membranes. Biochemistry 1981, 20, 7366–7373. (51) Altenbach, C.; Seelig, J. Calcium binding to phosphatidylcholine bilayers as studied by deuterium magnetic resonance. Evidence for the formation of a calcium complex with two phospholipid molecules. Biochemistry 1984, 23, 3913–3920. (52) Roux, M.; Bloom, M. Calcium, magnesium, lithium, sodium, and potassium distributions in the headgroup region of binary membranes of phosphatidylcholine and phosphatidylserine as seen by deuterium NMR. Biochemistry 1990, 29, 7077–7089. (53) Roux, M.; Bloom, M. Calcium binding by phosphatidylserine headgroups. Deuterium {NMR} study. Biophys. J. 1991, 60, 38 – 44. (54) Gally, H. U.; Niederberger, W.; Seelig, J. Conformation and motion of the choline head group in bilayers of dipalmitoyl-3-sn-phosphatidylcholine. Biochemistry 1975, 14, 3647–3652. (55) Scherer, P. G.; Seelig, J. Electric charge effects on phospholipid headgroups. Phosphatidylcholine in mixtures with cationic and anionic amphiphiles. Biochemistry 1989, 28, 7720–7728. (56) Browning, J. L.; Akutsu, H. Local anesthetics and divalent cations have the same effect

36

ACS Paragon Plus Environment

Page 36 of 48

Page 37 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

on the headgroups of phosphatidylcholine and phosphatidylethanolamine. Biochim. Biophys. Acta 1982, 684, 172 – 178. (57) Kelusky, E. C.; Smith, I. C. The influence of local anesthetics on molecular organization in phosphatidylethanolamine membranes. Mol. Pharmacol. 1984, 26, 314–321. (58) Roux, M.; Neumann, J. M.; Hodges, R. S.; Devaux, P. F.; Bloom, M. Conformational changes of phospholipid headgroups induced by a cationic integral membrane peptide as seen by deuterium magnetic resonance. Biochemistry 1989, 28, 2313–2321. (59) Kuchinka, E.; Seelig, J. Interaction of melittin with phosphatidylcholine membranes. Binding isotherm and lipid head-group conformation. Biochemistry 1989, 28, 4216– 4221. (60) Catte, A.; Girych, M.; Javanainen, M.; Miettinen, M. S.; Monticelli, L.; M¨a¨att¨a, J.; Oganesyan, V. S.; Ollila, O. H. S. The electrometer concept and binding of cations to phospholipid bilayers. 2015; DOI: 10.5281/zenodo.32175. (61) The NMRLipids collaboration,; Miettinen, S. M.; Ollila, O. H. S.; Botan, A.; Catte, A.; Edholm, O.; Favela, F.; Ferreira, T.; Fuchs, P.; Girych, M. et al. The NMRLipids project. 2015; DOI: 10.6084/m9.figshare.1585017. (62) Gowers, T.; Nielsen, M. Massively collaborative mathematics. Nature 2009, 461, 879– 881. (63) Ollila, O. H. S. Response of the hydrophilic part of lipid membranes to changing conditions - a critical comparison of simulations to experiments. 2013; http://arxiv. org/abs/1309.2131v1. (64) The NMRLipids collaboration,; Ollila, S.; Miettinen, S. M. On credits. 2015; DOI: 10.6084/m9.figshare.1577577.

37

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(65) Botan, A.; Favela-Rosales, F.; Fuchs, P. F. J.; Javanainen, M.; Kanduˆc, M.; Kulig, W.; Lamberg, A.; Loison, C.; Lyubartsev, A.; Miettinen, M. S. et al. nmrlipids.blogspot.fi: Final submission to the J. Phys. Chem. B. 2015; DOI: 10.5281/zenodo.32689. (66) Seelig, J. Deuterium magnetic resonance: theory and application to lipid membranes. Q. Rev. Biophys. 1977, 10, 353–418. (67) Hong, M.; Schmidt-Rohr, K.; Pines, A. NMR Measurement of Signs and Magnitudes of C-H Dipolar Couplings in Lecithin. J. Am. Chem. Soc. 1995, 117, 3310–3311. (68) Gross, J. D.; Warschawski, D. E.; Griffin, R. G. Dipolar Recoupling in MAS NMR: A Probe for Segmental Order in Lipid Bilayers. J. Am. Chem. Soc. 1997, 119, 796–802. (69) Dvinskikh, S. V.; Castro, V.; Sandstrom, D. Efficient solid-state NMR methods for measuring heteronuclear dipolar couplings in unoriented lipid membrane systems. Phys. Chem. Chem. Phys. 2005, 7, 607–613. (70) The NMRLipids collaboration,; Ollila, S.; Miettinen, S. M.; Vogel, A. Accuracy of order parameter measurements. 2015; DOI: 10.6084/m9.figshare.1577576. (71) The NMRLipids collaboration,; Ollila, S.; Fuchs, P.; Javanainen, M.; Lamberg, A. On the signs of the order parameters. 2015; DOI: 10.6084/m9.figshare.1577578. (72) Engel, A. K.; Cowburn, D. The origin of multiple quadrupole couplings in the deuterium {NMR} spectra of the 2 chain of 1,2 dipalmitoyl-sn-glycero-3phosphorylcholine. FEBS Letters 1981, 126, 169 – 171. (73) Vogel, A.; Feller, S. Headgroup Conformations of Phospholipids from Molecular Dynamics Simulation: Sampling Challenges and Comparison to Experiment. The Journal of Membrane Biology 2012, 245, 23–28. (74) Ollila, S.; Hyv¨onen, M. T.; Vattulainen, I. Polyunsaturation in Lipid Membranes:

38

ACS Paragon Plus Environment

Page 38 of 48

Page 39 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Dynamic Properties and Lateral Pressure Profiles. J. Phys. Chem. B 2007, 111, 3139– 3150. (75) Ferreira, T. M.; Ollila, O. H. S.; Pigliapochi, R.; Dabkowska, A. P.; Topgaard, D. Model-free estimation of the effective correlation time for CH bond reorientation in amphiphilic bilayers: 1H13C solid-state NMR and MD simulations. J. Chem. Phys. 2015, 142, 044905. (76) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. J. Chem. Theory Comput. 2008, 4, 435–447. (77) Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117, 1 – 19. (78) Lyubartsev, A. P.; Laaksonen, A. M.DynaMix a scalable portable parallel {MD} simulation package for arbitrary molecular mixtures. Comp. Phys. Comm. 2000, 128, 565 – 589. (79) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kal´e, L.; Schulten, K. Scalable molecular dynamics with NAMD. J. Comput. Chem. 2005, 26, 1781–1802. (80) Gurtovenko, A. A.;

Patra, M.;

Karttunen, M.;

Vattulainen, I. Cationic

DMPC/DMTAP Lipid Bilayers: Molecular Dynamics Study. Biophys. J. 2004, 86, 3461 – 3472. (81) Miettinen, M. S. Molecular dynamics simulation trajectory of a fully hydrated DMPC lipid bilayer. 2013; DOI: 10.6084/m9.figshare.829642. (82) Miettinen, M. S.; Gurtovenko, A. A.; Vattulainen, I.; Karttunen, M. Ion Dynamics

39

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

in Cationic Lipid Bilayer Systems in Saline Solutions. J. Phys. Chem. B 2009, 113, 9226–9234. (83) Marrink, S.-J.; Berger, O.; Tieleman, P.; J¨ahnig, F. Adhesion Forces of Lipids in a Phospholipid Membrane Studied by Molecular Dynamics Simulations. Biophys. J. 1998, 74, 931 – 943. (84) M¨aa¨tt¨a, J. DPPC Berger. 2015; DOI: 10.5281/zenodo.13934. (85) Ollila, O. H. S.; Ferreira, T.; Topgaard, D. MD simulation trajectory and related files for POPC bilayer (Berger model delivered by Tieleman, Gromacs 4.5). 2014; DOI: 10.5281/zenodo.13279. (86) Ollila, O. H. S.; Miettinen, M. MD simulation trajectory and related files for DPPC bilayer (CHARMM36, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.15549. (87) Ollila, O. H. S.; Miettinen, M. MD simulation trajectory and related files for POPC bilayer (CHARMM36, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.13944. (88) Santuz, H. MD simulation trajectory and related files for POPC bilayer (CHARMM36, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.14066. (89) Kulig, W.; Pasenkiewicz-Gierula, M.; R´og, T. Cis and Trans Unsaturated Phosphatidylcholine Bilayers: A Molecular Dynamics Simulation Study. Chem. Phys. Lipids 2015, DOI: 10.1016/j.chemphyslip.2015.07.002. (90) Javanainen, M. POPC @ 310K, model by Maciejewski and Rog. 2014; DOI: 10.5281/zenodo.13497. (91) Javanainen, M. POPC/Cholesterol @ 310K. 0, 10, 40, 50 and 60 mol-cholesterol. Model by Maciejewski and Rog. 2015; DOI: 10.5281/zenodo.13877. (92) Javanainen, M. POPC @ 310K, varying water-to-lipid ratio. Model by Maciejewski and Rog. 2014; DOI: 10.5281/zenodo.13498. 40

ACS Paragon Plus Environment

Page 40 of 48

Page 41 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(93) Ollila, O. H. S.; Retegan, M. MD simulation trajectory and related files for DPPC bilayer (GAFFlipid, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.15550. (94) Ollila, O. H. S.; Retegan, M. MD simulation trajectory and related files for POPC bilayer (GAFFlipid, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.13791. (95) Dickson, C. J.; Madej, B. D.; Skjevik, . A.; Betz, R. M.; Teigen, K.; Gould, I. R.; Walker, R. C. Lipid14: The Amber Lipid Force Field. J. Chem. Theory Comput. 2014, 10, 865–879. (96) Ollila, O. H. S.; Retegan, M. MD simulation trajectory and related files for POPC bilayer (Lipid14, Gromacs 4.5). 2014; DOI: 10.5281/zenodo.12767. (97) Poger, D.; Van Gunsteren, W. F.; Mark, A. E. A new force field for simulating phosphatidylcholine bilayers. J. Comput. Chem. 2010, 31, 1117–1125. (98) Fuchs, P. F. MD simulation trajectory and related files for DPPC bilayer in full hydration (Poger GROMOS53A6 L, Gromacs 4.0.7, PME, traj 1). 2015; DOI: 10.5281/zenodo.14594. (99) Fuchs, P. F. MD simulation trajectory and related files for DPPC bilayer in full hydration (Poger GROMOS53A6 L, Gromacs 4.0.7, PME, traj 2). 2015; DOI: 10.5281/zenodo.14595. (100) J¨ambeck, J. P. M.; Lyubartsev, A. P. Derivation and Systematic Validation of a Refined All-Atom Force Field for Phosphatidylcholine Lipids. J. Phys. Chem. B 2012, 116, 3164–3179. (101) M¨aa¨tt¨a, J. DPPC Slipids. 2014; DOI: 10.5281/zenodo.13287. (102) J¨ambeck, J. P. M.; Lyubartsev, A. P. An Extension and Further Validation of an AllAtomistic Force Field for Biological Membranes. J. Chem. Theory Comput. 2012, 8, 2938–2948. 41

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(103) Javanainen, M. POPC @ 310K, Slipids force field. 2015; DOI: 10.5281/zenodo.13887. (104) Kukol, A. Lipid Models for United-Atom Molecular Dynamics Simulations of Proteins. J. Chem. Theory Comput. 2009, 5, 615–626. (105) Javanainen, M. POPC @ 298K, Model by Kukol. 2014; DOI: 10.5281/zenodo.13393. (106) Chiu, S.-W.; Pandit, S. A.; Scott, H. L.; Jakobsson, E. An Improved United Atom Force Field for Simulation of Mixed Lipid Bilayers. J. Phys. Chem. B 2009, 113, 2748–2763. (107) Ollila, O. H. S. MD simulation trajectory and related files for POPC bilayer (Chiu et al. Gromos version, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.15548. (108) Lyubartsev, A. MD simulation trajectory and related files for DMPC bilayer, H¨ogberg et al, J.Comp.Chem., 29, 2359 (2008). 2015; DOI: 10.5281/zenodo.16195. (109) Rabinovich, A. L.; Lyubartsev, A. P. Bond orientation properties in lipid molecules of membranes: molecular dynamics simulations. J. Phys. Conf. Ser. 2014, 510, 012022. (110) Lyubartsev, A. MD simulation trajectory and related files for POPC bilayer, H¨ogberg et al parameters (J.Comp.Chem., 29, 2359 (2008)). 2015; DOI: 10.5281/zenodo.16724. (111) Ulmschneider, J. P.; Ulmschneider, M. B. United Atom Lipid Parameters for Combination with the Optimized Potentials for Liquid Simulations All-Atom Force Field. J. Chem. Theory Comput. 2009, 5, 1803–1813. (112) Javanainen, M. POPC @ 310K, Model by Ulmschneider and Ulmschneider. 2014; DOI: 10.5281/zenodo.13392. (113) Tj¨ornhammar, R.; Edholm, O. Reparameterized United Atom Model for Molecular Dynamics Simulations of Gel and Fluid Phosphatidylcholine Bilayers. J. Chem. Theory Comput. 2014, 10, 5706–5715. 42

ACS Paragon Plus Environment

Page 42 of 48

Page 43 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(114) Javanainen, M. DPPC @ 323K, new FF by Tj¨ornhammar and Edholm. 2014; DOI: 10.5281/zenodo.12743. (115) Henin, J.; Shinoda, W.; Klein, M. L. United-Atom Acyl Chains for CHARMM Phospholipids. J. Phys. Chem. B 2008, 112, 7008–7015. (116) Botan, A. DLPC@ 323K, CHARMM36UA force field. 2015; DOI: 10.5281/zenodo.13821. (117) Lee, S.; Tran, A.; Allsopp, M.; Lim, J. B.; Henin, J.; Klauda, J. B. CHARMM36 United Atom Chain Model for Lipids and Surfactants. J. Phys. Chem. B 2014, 118, 547–556. (118) Loison, C. Hydrated DPPC, MD simulation trajectory and related files for UA charmm36 model by Lee et al 2014. 2015; DOI: 10.5281/zenodo.17004. (119) Ollila, O. H. S. MD simulation trajectory and related files for POPC bilayer in low hydration (Berger model delivered by Tieleman, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.13814. (120) Kanduc, M.; Schneck, E.; Netz, R. R. Hydration Interaction between Phospholipid Membranes: Insight into Different Measurement Ensembles from Atomistic Molecular Dynamics Simulations. Langmuir 2013, 29, 9126–9137. (121) Kanduc, M. MD trajectory for DLPC bilayer (Berger, Gromacs 4.5.4), nw=28 w/l. 2015; DOI: 10.5281/zenodo.16287. (122) Kanduc, M. MD trajectory for DLPC bilayer (Berger, Gromacs 4.5.4), nw=24 w/l. 2015; DOI: 10.5281/zenodo.16289. (123) Kanduc, M. MD trajectory for DLPC bilayer (Berger, Gromacs 4.5.4), nw=20 w/l. 2015; DOI: 10.5281/zenodo.16291.

43

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(124) Kanduc, M. MD trajectory for DLPC bilayer (Berger, Gromacs 4.5.4), nw=16 w/l. 2015; DOI: 10.5281/zenodo.16292. (125) Kanduc, M. MD trajectory for DLPC bilayer (Berger, Gromacs 4.5.4), nw=12 w/l. 2015; DOI: 10.5281/zenodo.16293. (126) Kanduc, M. MD trajectory for DLPC bilayer (Berger, Gromacs 4.5.4), nw=8 w/l. 2015; DOI: 10.5281/zenodo.16294. (127) Kanduc, M. MD trajectory for DLPC bilayer (Berger, Gromacs 4.5.4), nw=4 w/l. 2015; DOI: 10.5281/zenodo.16295. (128) Ollila, O. H. S.; Miettinen, M. MD simulation trajectory and related files for POPC bilayer in medium low hydration (CHARMM36, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.13946. (129) Ollila, O. H. S.; Miettinen, M. MD simulation trajectory and related files for POPC bilayer in low hydration (CHARMM36, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.13945. (130) Ollila, O. H. S. MD simulation trajectory and related files for POPC bilayer in low hydration (GAFFlipid, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.13853. (131) H¨oltje, M.; F¨orster, T.; Brandt, B.; Engels, T.; von Rybinski, W.; H¨oltje, H.-D. Molecular dynamics simulations of stratum corneum lipid models: fatty acids and cholesterol. Biochim. Biophys. Acta 2001, 1511, 156 – 167. (132) Ollila, O. H. S.; Ferreira, T.; Topgaard, D. MD simulation trajectory and related files for POPC/cholesterol (7 mol%) bilayer (Berger model delivered by Tieleman, modified H¨oltje, Gromacs 4.5). 2014; DOI: 10.5281/zenodo.13282. (133) Ollila, O. H. S.; Ferreira, T.; Topgaard, D. MD simulation trajectory and related

44

ACS Paragon Plus Environment

Page 44 of 48

Page 45 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

files for POPC/cholesterol (15 mol%) bilayer (Berger model delivered by Tieleman, modified H¨oltje, Gromacs 4.5). 2014; DOI: 10.5281/zenodo.13281. (134) Ollila, O. H. S.; Ferreira, T.; Topgaard, D. MD simulation trajectory and related files for POPC/cholesterol (34 mol%) bilayer (Berger model delivered by Tieleman, modified H¨oltje, Gromacs 4.5). 2014; DOI: 10.5281/zenodo.13283. (135) Ollila, O. H. S.; Ferreira, T.; Topgaard, D. MD simulation trajectory and related files for POPC/cholesterol (50 mol%) bilayer (Berger model delivered by Tieleman, modified H¨oltje, Gromacs 4.5). 2014; DOI: 10.5281/zenodo.13285. (136) Ollila, O. H. S.; Ferreira, T.; Topgaard, D. MD simulation trajectory and related files for POPC/cholesterol (60 mol%) bilayer (Berger model delivered by Tieleman, modified H¨oltje, Gromacs 4.5). 2014; DOI: 10.5281/zenodo.13286. (137) Lim, J. B.; Rogaski, B.; Klauda, J. B. Update of the Cholesterol Force Field Parameters in CHARMM. J. Phys. Chem. B 2012, 116, 203–210. (138) Favela-Rosales, F. POPC Glycerol CHARMM36 0-10-15-20-25-35-50%CHOL. 2015; DOI: 10.5281/zenodo.16830. (139) Santuz, H. MD simulation trajectory for POPC/20% Chol bilayer (CHARMM36, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.14067. (140) Santuz, H. MD simulation trajectory for POPC/50% Chol bilayer (CHARMM36, Gromacs 4.5). 2015; DOI: 10.5281/zenodo.14068. (141) Warschawski, D.; Devaux, P. Order parameters of unsaturated phospholipids in membranes and the effect of cholesterol: a 1H13C solid-state NMR study at natural abundance. Eur. Biophys. J. 2005, 34, 987–996. (142) Mashl, R. J.; Scott, H. L.; Subramaniam, S.; Jakobsson, E. Molecular Simulation of

45

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Dioleoylphosphatidylcholine Lipid Bilayers at Differing Levels of Hydration. Biophys. J. 2001, 81, 3005 – 3015. (143) Pertsin, A.; Platonov, D.; Grunze, M. Direct computer simulation of water-mediated force between supported phospholipid membranes. J. Chem. Phys. 2005, 122, 244708. (144) Pertsin, A.; Platonov, D.; Grunze, M. Origin of Short-Range Repulsion between Hydrated Phospholipid Bilayers: A Computer Simulation Study. Langmuir 2007, 23, 1388–1393. (145) Eun, C.; Berkowitz, M. L. Origin of the Hydration Force: Water-Mediated Interaction between Two Hydrophilic Plates. J. Phys. Chem. B 2009, 113, 13222–13228. (146) Eun, C.; Berkowitz, M. L. Thermodynamic and Hydrogen-Bonding Analyses of the Interaction between Model Lipid Bilayers. J. Phys. Chem. B 2010, 114, 3013–3019. (147) Schneck, E.; Sedlmeier, F.; Netz, R. R. Hydration repulsion between biomembranes results from an interplay of dehydration and depolarization. Proc. Natl. Acad. Sci. USA 2012, 109, 14405–14409. (148) Israelachvili, J. N. Intermolecular and Surface Forces; Academic Press: London, 1985. (149) Israelachvili, J. N.; Wennerstr¨om, H. Role of hydration and water structure in biological and colloidal interactions. Nature 1996, 379, 219 – 225. (150) Sparr, E.; Wennerstr¨om, H. Interlamellar forces and the thermodynamic characterization of lamellar phospholipid systems. Curr. Opin. Colloid Interf. Science 2011, 16, 561 – 567. (151) Simons, K.; Vaz, W. L. Model Systems, Lipid Rafts, And Cell Membranes. Ann. Rev. Biophys. Biomol. Struct. 2004, 33, 269–295.

46

ACS Paragon Plus Environment

Page 46 of 48

Page 47 of 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(152) Somerharju, P.; Virtanen, J. A.; Cheng, K. H.; Hermansson, M. The superlattice model of lateral organization of membranes and its implications on membrane lipid homeostasis. Biochim. Biophys. Acta - Biomembranes 2009, 1788, 12 – 23. (153) Huang, J.; Feigenson, G. W. A Microscopic Interaction Model of Maximum Solubility of Cholesterol in Lipid Bilayers. Biophys. J. 1999, 76, 2142 – 2157. (154) Zhu, Q.; Cheng, K. H.; Vaughn, M. W. Molecular Dynamics Studies of the Molecular Structure and Interactions of Cholesterol Superlattices and Random Domains in an Unsaturated Phosphatidylcholine Bilayer Membrane. J. Phys. Chem. B 2007, 111, 11021–11031. (155) Rog, T.; Pasenkiewicz-Gierula, M.; Vattulainen, I.; Karttunen, M. Ordering effects of cholesterol and its analogues. Biochim. Biophys. Acta 2009, 1788, 97 – 121. (156) Alwarawrah, M.; Dai, J.; Huang, J. Modification of Lipid Bilayer Structure by Diacylglycerol: A Comparative Study of Diacylglycerol and Cholesterol. J. Chem. Theor. Comput. 2012, 8, 749–758. (157) Marsh, D. Liquid-ordered phases induced by cholesterol: A compendium of binary phase diagrams. Biochim. Biophys. Acta 2010, 1798, 688 – 699. (158) Ghosh, R.;

Seelig, J. The interaction of cholesterol with bilayers of phos-

phatidylethanolamine. Biochim. Biophys. Acta 1982, 691, 151 – 160. (159) Ferreira, T. M.; Topgaard, D.; Ollila, O. H. S. O. Molecular Conformation and Bilayer Pores in a Nonionic Surfactant Lamellar Phase Studied with 1H13C Solid-State NMR and Molecular Dynamics Simulations. Langmuir 2014, 30, 461–469.

47

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 48 of 48